With over 10 million monthly active merchant users, Khatabook has become one of the notable household names for kiranas and mom-and-pop stores across the country. Interestingly, many of the users using Khatabook apps are first-time internet users, meaning they have no previous digital footprint. 

Khatabook’s data science team uses the new insights they get from its users every day to understand user behaviour, mitigate payment and credit risk, forecast adoption and business growth, and lead generation and prioritisation. 

The company leverages data science to identify the nature of merchant users, whether they are suppliers or retailers, alongside the category of business they operate in, the primary purpose they are maintaining Khatabook for, etc. It also uses this to decipher strategy and evaluate product performance. 

When it comes to mitigating risk, Khatabook’s data science team uses advanced artificial intelligence (AI) and machine learning (ML) models to mitigate risk. Data science is also the core engine behind credit underwriting to merchants. 

Business predictability is another area where Khatabook’s data science team helps build the business decision. “We estimate the organic and paid customer growth six months in advance based on the data science framework, considering all other external factors, new user acquisition, old user retention aspects, marketing spends, and other important variables,” shared Ravish Naresh, CEO and co-founder at Khatabook. 

Currently, KhataBook has five different products – namely Gold Rate, Payments, Business Tips, GST, Salary – thus making a compelling use case of cross-selling and upselling. Cross-selling and upselling are usually performed in the mid-to-late stage of the conversion funnel, where a customer has already indicated they are likely to buy a product or service.

Khatabook uses data science to identify potential users of other apps and identify who will be more likely to purchase the premium plans. “All the cross-selling and lead conversion efforts are prioritised using data science models,” added Naresh. 

Further, he said, at Khatabook, they handle risk by using a mix of auto-blockers that block suspicious users based on some preset rules and an ML model which gives a risk score for potential fraud users. 

“The auto-blockers and ML-based risk model work in tandem to prevent fraud within the Khatabook payment ecosystem. Our CTS (cheque truncation system) and FTS (fraud to sales) numbers have remained healthy and way below the industry cutoffs,” said Naresh. 

Expanding Mode: ON 

The team told Analytics India Magazine that it is currently hiring for various data science roles, including director of data science, associate director of analytics, manager analytics, and senior data scientist positions. The company is looking for candidates with anywhere between three to eight years of experience. 

Khatabook is hiring data science and machine learning professionals to address three key objectives – 

• Identifying the authenticity of merchants for fraud risk and lending 
• Forecasting the business performance and planning accordingly 
• Make the product personalised and build AI products to reduce the effort 

Team Structure 

At Khatabook, the analytics and data science team consists of 30 members. 

Analytics and data science is a centralised function in Khatabook. For instance, Khatabook has five products, and each product has a data science leader and a team of 4-5 data analysts/scientists. 

Interview Process 

The data science function in Khatabook has two streams – business analytics and machine learning.

So, the interview process for business analytics would include – 

• Screening round (technical analysis + business case study) – Knockout round 
• Another technical analysis round 
• Analytics case study round 
• Business head round 
• Cultural fitment round

For machine learning, the interview process would include: 

• Screening round (past projects + ML concepts) – Knockout round 
• Technical analysis round – machine learning 
• Business case study round 
• ML technical assignment (hands-on coding) 
• Cultural fitment round 

Are you cut out for this job? 

Here are some of the prerequisites to apply for data science and machine learning jobs at Khatabook. 

• Knowledge of various machine learning techniques (clustering, time series, decision tree learning, fraud/anomaly detection, recommendation algorithm, etc.) 
• Knowledge of advanced statistical techniques and concepts like regression, statistical tests and proper usage, properties of distributions, etc., and experience with applications
• Excellent communication skills (both written and verbal) for coordinating across teams 
• A drive to learn and master new technologies and techniques

Skills

Some of the skills required to work at Khatabook include 

• Technical capabilities: Proficient in SQL, Excel, other scripting languages (Python, R, etc.) 
• Strong working experience with BI tools (Tableau, Power BI, etc.) 

Data Science Tools

Khatabook uses tools like Git, Airflow, Tensorflow, Keras, MLOps, Mix panel, Snowflake, Tableau, etc. 

Expectations 

According to Khatabook, the ideal candidate should demonstrate four core capabilities: 

• Good business acumen and structured problem-solving skills 
• Impeccable product sense 
• Technical expertise in machine learning/analytics techniques 
• High sense of ownership 

Besides these, some of the other parameters (KRAs and KPAs) used to assess candidates for the data science team at Khatabook include 

• Understanding business applications behind ML/analytics techniques 
• A strong foundation of statistics or ML algorithms applicable to roles 
• Bias for action; focus on business value generated out of insights 
• Writing clean and optimised production-ready code

Dos & Don’ts

Naresh said that most candidates who apply for data science jobs at Khatabook tend to give more weightage to explaining the models rather than understanding the problem. Plus, he said they do not give emphasis to presentation skills or are less articulate while explaining the assignment. “Some of them have a breadth of knowledge but miss out on the depth. It’s better to know just two techniques but know them in-depth,” added Naresh.  

Work Culture 

Naresh said that they have a lean team with higher responsibilities. The team is also horizontal and serves all business functions wherever opportunities lie. This provides the team with the freedom to work in every domain and solve a variety of problems. 

“Besides the amazing work lineup, we believe in ownership and interpersonal learning. Team members are free to design their own workstream by aligning them to the large objective,” said Naresh. “We encourage a lot of interpersonal learning and brainstorming with peers as we understand that we are dealing with very unique problems, and different perspectives will help us achieve better results.”

What can you expect from Khatabook? 

Khatabook is in the domain of tech for emerging markets. Many of the users have no previous digital footprint. With the increase in digital adoption, more new data is being generated, providing a fantastic opportunity for the team to work on exciting use cases and problem-solving. Also, user behaviour in emerging markets is evolving rapidly. This makes Khatabook an exciting place for any data science professional to work. 

In addition to this, Khatabook offers a plethora of work benefits to employees, such as unlimited leaves, extended paternity leaves, week-long company-wide mental health breaks at regular intervals, no meetings Wednesdays, referral bonuses, internet and home office expense claims, virtual team feasts covering, vaccine drives for employees, and virtual team cultural events translating into a solid work culture and collaborative team environment. 

Are you ready? 

Lastly, here are some of the key things to keep in mind when applying for data science roles at Khatabook – 

• Be prepared to answer the business value of your work 
• Brush up the concepts behind machine learning techniques you know and also understand their wider business applications 
• Learn how to work with unstructured problems; solve case studies 
• Have a good idea of data engineering concepts to implement models in production

So, what are you waiting for? Click here to apply for data science jobs at Khatabook today!